Meta's Universal Translator Will Not Be the End of Human Translation

Will Meta’s Universal Translator be the end of human translations in the translation agency? This article examines why it will be challenging to create one.

For those who grew up watching Sci-Fi shows, like Star Trek, you must have remembered the Universal Translator? It was one of the most interesting technologies featured in the show as it allowed Captain Kirk’s team to understand the Alien language of the natives that they encountered in real-time.

We have pocket translators and online translation tools in real life, any translation agency would know that they are prone to mistranslation even for simple phrases. But that hasn’t stopped Mark Zuckerberg, the owner of Meta, which manages social media giants like Facebook, and Instagram, to start developing an AI-powered technology that can help us communicate with anyone regardless of their language.

Today, we’ll discuss the Universal Translator, the current technology developed by the translation agency, and why this technology won’t spell the end of human translation. To learn more about this, keep on reading!

The universal translator of Meta

To begin our discussion on Meta’s universal translator, let’s first discuss what it’s all about. According to Meta’s blog, it’s a translation-based initiative that aims to build AI programs that have a standardized writing system that will allow them to learn how to translate from one language to another without needing many examples for its training.

It also aims to develop translation speech in real-time without relying on writing it down, which is what most translation apps are doing.

This initiative is made due to the rise of Virtual Reality (VR) and Augmented Reality (AR) platforms, as it would “break down” these barriers. However, this long-term project by Meta currently has no clear timeframe or roadmap in reaching its goal. This has sparked a lot of discussion amongst the translation services agency regarding the future of their industry.

The translation industry’s technological advancement over the years

The first successful translation technology occurred in the 1950s, a collaborative effort of IBM and Georgetown University. It had a very limited range of vocabulary, as it could only translate about 250 words from Russian to English.

Since then, translation technology has drastically improved to the point that it aids almost any translation services agency in providing accurate and fast delivery to several industries.

This technology has influenced the translation industry and other sectors in society, but it has also helped every internet user. GWI reported that one-third of working-age online users utilize online translation tools every week to translate text. But they said this figure could be higher in some locations.

The term AI originally referred to any technology taking up the translation process by the 1956 Dartmouth Conference. Meta’s attempt to create a Universal Translator, although they’re not the first one to start building one, will be the first one with the purpose that it will be used in an online interactive platform in real-time.

Despite the advancement of AI in over a decade, it’s still much in its infancy. In turn, it affects the development of translation technology. However, the fact they’re beginning to create an AI system that translates speech in real without having to convert it into text is a commendable effort.

What a universal translator means for the virtual world

As mentioned, the development of a universal translator is due to the rise of VR and AR platforms. Meta has acknowledged that although the most commonly spoken languages, such as English, Spanish, and Mandarin, are available in several translation tools, about 20% of the entire world’s population are left behind in this respect.

Due to this, oral communication in the virtual world will become important as it allows a more immersive experience for users. Currently, most speech-based translation technology follows a speech-to-text system before it can be translated. This delays the translation process and increases the chances of miscommunication.

As of now, many automated translations used online and by professional translation agencies follow the statistical model that relies on syntax and format of the text and the machine’s experience with certain language pairs. Due to these limitations, it’s no wonder that Meta is creating a virtual interpretation device that will allow speech-to-speech translation immediately without going through a third platform.

What a universal translator means in the real world

Meta’s announcement in creating a universal translator has stirred up a lot of discussion among professional translation agencies. If successfully done, it would allow individuals who have access to Meta’s platforms to communicate with users worldwide, just like the characters in Star Trek.

It could also improve the translation agency’s translation apps and devices because, as mentioned, most of them still rely on speech-to-text translation frameworks. Incorporating an AI framework into the universal translator could revolutionize the way we communicate as we wouldn’t have to rely on an interpreter to translate what we say.

But this is all theoretical, as the application in implementing AI to create speech-to-speech translation will take time.

Meta’s present translation technology, the same one used by Facebook and Instagram, has been criticized for its inaccuracy. An example of how severely problematic it can be, occurred in 2017 when the Israeli police arrested a Palestinian due to Facebook’s automated translation mistranslating his Arabic Facebook post that meant “good morning” to “hurt them” in English and “attack them” in Hebrew.

Meta can ensure their translation software’s quality by having organizations and associations evaluate them, like the International Organization for Standardization (ISO). A translation agency that utilizes machine translation post-editing (MTPE) obtains and gets certified for ISO 18587:2017 to help determine their translations’ quality. It also evaluates the skills of their post-edited specialists and linguists working with these technologies.

Why universal translators won’t end human translation

Even though Mark Zuckerberg believes that we will have a universal translator in our lifetime through AI is a tough feat. It’s because language is ever-evolving. The meaning of specific words and phrases can change from generation to generation. Because a generation has a total span of 20 to 30 years, there’s a problem of technology not being able to catch up with the changes in grammar, slang, definitions, and accepted words in the dictionary.

One of the challenges that big tech companies face when it comes to creating translation software is that even though a language is used across different countries, one language creates several variations as it is adapted in each country.

An example of this is Hispanic countries. In most Spanish-speaking countries, the English word ‘baby’ is ‘bebe.’ However, in Chile, ‘guagua’ means baby, while ‘guagua’ means bus in Cuba. And so, considering the cultural nuance of any target country is essential as you could easily offend the people you’re trying to attract, which translation technology still lacks.

The future of professional translation agencies

The ones mentioned above are some of the many issues and challenges Meta will encounter in creating a universal translator. This is also why many professional translation agencies still have linguists manage projects and why we shouldn’t be worried about machines taking over the translation industry any time soon.

Regardless of the development of Meta’s speech-to-speech translation software, the most probable outcome is that linguists will still be needed to facilitate the virtual world’s communication. Because despite its completion through machine learning technology, it’s not guaranteed that the machines will have enough level of cultural nuance to provide accurate translations. Over-reliance on the universal translator for communication could cause problems like what happened to the Palestinians who just posted “good morning” on Facebook.

Conclusion

We still have a long way to go before we can achieve a fully functioning universal translator, and only time will tell if we will be carrying devices like the Star Trek crew when we travel in the real and virtual world within the coming years. So far, the biggest problem that Meta will have to resolve is ensuring the quality of the translation, which will most likely end with human translators still managing the technology like what the translation agency is currently doing.